renames_test.go 9.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308
  1. package plumbing
  2. import (
  3. "bytes"
  4. "compress/gzip"
  5. "io/ioutil"
  6. "os"
  7. "path"
  8. "testing"
  9. "github.com/stretchr/testify/assert"
  10. "gopkg.in/src-d/go-git.v4/plumbing"
  11. "gopkg.in/src-d/go-git.v4/plumbing/object"
  12. "gopkg.in/src-d/hercules.v9/internal/core"
  13. "gopkg.in/src-d/hercules.v9/internal/test"
  14. )
  15. func fixtureRenameAnalysis() *RenameAnalysis {
  16. ra := RenameAnalysis{SimilarityThreshold: 80}
  17. ra.Initialize(test.Repository)
  18. return &ra
  19. }
  20. func TestRenameAnalysisMeta(t *testing.T) {
  21. ra := fixtureRenameAnalysis()
  22. assert.Equal(t, ra.Name(), "RenameAnalysis")
  23. assert.Equal(t, len(ra.Provides()), 1)
  24. assert.Equal(t, ra.Provides()[0], DependencyTreeChanges)
  25. assert.Equal(t, len(ra.Requires()), 2)
  26. assert.Equal(t, ra.Requires()[0], DependencyBlobCache)
  27. assert.Equal(t, ra.Requires()[1], DependencyTreeChanges)
  28. opts := ra.ListConfigurationOptions()
  29. assert.Len(t, opts, 1)
  30. assert.Equal(t, opts[0].Name, ConfigRenameAnalysisSimilarityThreshold)
  31. ra.SimilarityThreshold = 0
  32. facts := map[string]interface{}{}
  33. facts[ConfigRenameAnalysisSimilarityThreshold] = 70
  34. ra.Configure(facts)
  35. assert.Equal(t, ra.SimilarityThreshold, 70)
  36. delete(facts, ConfigRenameAnalysisSimilarityThreshold)
  37. ra.Configure(facts)
  38. assert.Equal(t, ra.SimilarityThreshold, 70)
  39. }
  40. func TestRenameAnalysisRegistration(t *testing.T) {
  41. summoned := core.Registry.Summon((&RenameAnalysis{}).Name())
  42. assert.Len(t, summoned, 1)
  43. assert.Equal(t, summoned[0].Name(), "RenameAnalysis")
  44. summoned = core.Registry.Summon((&RenameAnalysis{}).Provides()[0])
  45. assert.True(t, len(summoned) >= 1)
  46. matched := false
  47. for _, tp := range summoned {
  48. matched = matched || tp.Name() == "RenameAnalysis"
  49. }
  50. assert.True(t, matched)
  51. }
  52. func TestRenameAnalysisInitializeInvalidThreshold(t *testing.T) {
  53. ra := RenameAnalysis{SimilarityThreshold: -10}
  54. ra.Initialize(test.Repository)
  55. assert.Equal(t, ra.SimilarityThreshold, RenameAnalysisDefaultThreshold)
  56. ra = RenameAnalysis{SimilarityThreshold: 110}
  57. ra.Initialize(test.Repository)
  58. assert.Equal(t, ra.SimilarityThreshold, RenameAnalysisDefaultThreshold)
  59. ra = RenameAnalysis{SimilarityThreshold: 0}
  60. ra.Initialize(test.Repository)
  61. ra = RenameAnalysis{SimilarityThreshold: 100}
  62. ra.Initialize(test.Repository)
  63. }
  64. func TestRenameAnalysisConsume(t *testing.T) {
  65. ra := fixtureRenameAnalysis()
  66. changes := make(object.Changes, 3)
  67. // 2b1ed978194a94edeabbca6de7ff3b5771d4d665
  68. treeFrom, _ := test.Repository.TreeObject(plumbing.NewHash(
  69. "96c6ece9b2f3c7c51b83516400d278dea5605100"))
  70. treeTo, _ := test.Repository.TreeObject(plumbing.NewHash(
  71. "251f2094d7b523d5bcc60e663b6cf38151bf8844"))
  72. changes[0] = &object.Change{From: object.ChangeEntry{
  73. Name: "analyser.go",
  74. Tree: treeFrom,
  75. TreeEntry: object.TreeEntry{
  76. Name: "analyser.go",
  77. Mode: 0100644,
  78. Hash: plumbing.NewHash("baa64828831d174f40140e4b3cfa77d1e917a2c1"),
  79. },
  80. }, To: object.ChangeEntry{},
  81. }
  82. changes[1] = &object.Change{From: object.ChangeEntry{}, To: object.ChangeEntry{
  83. Name: "burndown.go",
  84. Tree: treeTo,
  85. TreeEntry: object.TreeEntry{
  86. Name: "burndown.go",
  87. Mode: 0100644,
  88. Hash: plumbing.NewHash("29c9fafd6a2fae8cd20298c3f60115bc31a4c0f2"),
  89. },
  90. },
  91. }
  92. changes[2] = &object.Change{From: object.ChangeEntry{
  93. Name: "cmd/hercules/main.go",
  94. Tree: treeFrom,
  95. TreeEntry: object.TreeEntry{
  96. Name: "cmd/hercules/main.go",
  97. Mode: 0100644,
  98. Hash: plumbing.NewHash("c29112dbd697ad9b401333b80c18a63951bc18d9"),
  99. },
  100. }, To: object.ChangeEntry{
  101. Name: "cmd/hercules/main.go",
  102. Tree: treeTo,
  103. TreeEntry: object.TreeEntry{
  104. Name: "cmd/hercules/main.go",
  105. Mode: 0100644,
  106. Hash: plumbing.NewHash("f7d918ec500e2f925ecde79b51cc007bac27de72"),
  107. },
  108. },
  109. }
  110. cache := map[plumbing.Hash]*CachedBlob{}
  111. AddHash(t, cache, "baa64828831d174f40140e4b3cfa77d1e917a2c1")
  112. AddHash(t, cache, "29c9fafd6a2fae8cd20298c3f60115bc31a4c0f2")
  113. AddHash(t, cache, "c29112dbd697ad9b401333b80c18a63951bc18d9")
  114. AddHash(t, cache, "f7d918ec500e2f925ecde79b51cc007bac27de72")
  115. deps := map[string]interface{}{}
  116. deps[DependencyBlobCache] = cache
  117. deps[DependencyTreeChanges] = changes
  118. ra.SimilarityThreshold = 37
  119. res, err := ra.Consume(deps)
  120. assert.Nil(t, err)
  121. renamed := res[DependencyTreeChanges].(object.Changes)
  122. assert.Equal(t, len(renamed), 2)
  123. ra.SimilarityThreshold = 39
  124. res, err = ra.Consume(deps)
  125. assert.Nil(t, err)
  126. renamed = res[DependencyTreeChanges].(object.Changes)
  127. assert.Equal(t, len(renamed), 3)
  128. }
  129. func TestSortableChanges(t *testing.T) {
  130. changes := sortableChanges{
  131. sortableChange{
  132. nil, plumbing.NewHash("0000000000000000000000000000000000000000"),
  133. }, sortableChange{
  134. nil, plumbing.NewHash("ffffffffffffffffffffffffffffffffffffffff"),
  135. },
  136. }
  137. assert.True(t, changes.Less(0, 1))
  138. assert.False(t, changes.Less(1, 0))
  139. assert.False(t, changes.Less(0, 0))
  140. changes.Swap(0, 1)
  141. assert.Equal(t, changes[0].hash.String(), "ffffffffffffffffffffffffffffffffffffffff")
  142. assert.Equal(t, changes[1].hash.String(), "0000000000000000000000000000000000000000")
  143. }
  144. func TestSortableBlobs(t *testing.T) {
  145. blobs := sortableBlobs{
  146. sortableBlob{
  147. nil, int64(0),
  148. }, sortableBlob{
  149. nil, int64(1),
  150. },
  151. }
  152. assert.True(t, blobs.Less(0, 1))
  153. assert.False(t, blobs.Less(1, 0))
  154. assert.False(t, blobs.Less(0, 0))
  155. blobs.Swap(0, 1)
  156. assert.Equal(t, blobs[0].size, int64(1))
  157. assert.Equal(t, blobs[1].size, int64(0))
  158. }
  159. func TestRenameAnalysisFork(t *testing.T) {
  160. ra1 := fixtureRenameAnalysis()
  161. clones := ra1.Fork(1)
  162. assert.Len(t, clones, 1)
  163. ra2 := clones[0].(*RenameAnalysis)
  164. assert.True(t, ra1 == ra2)
  165. ra1.Merge([]core.PipelineItem{ra2})
  166. }
  167. func TestRenameAnalysisSizesAreClose(t *testing.T) {
  168. ra := fixtureRenameAnalysis()
  169. assert.True(t, ra.sizesAreClose(941, 963))
  170. assert.True(t, ra.sizesAreClose(941, 1150))
  171. assert.True(t, ra.sizesAreClose(941, 803))
  172. assert.False(t, ra.sizesAreClose(1320, 1668))
  173. }
  174. func TestRenameAnalysisSortRenameCandidates(t *testing.T) {
  175. candidates := []int{0, 1, 2, 3}
  176. sortRenameCandidates(candidates, "test_regression.py", func(i int) string {
  177. return []string{"gather_nd_op.h", "test.py", "test_file_system.cc", "regression.py"}[i]
  178. })
  179. assert.Equal(t, candidates[0], 3)
  180. assert.Equal(t, candidates[1], 1)
  181. }
  182. func TestBlobsAreCloseText(t *testing.T) {
  183. blob1 := &CachedBlob{Data: []byte("hello, world!")}
  184. blob2 := &CachedBlob{Data: []byte("hello, world?")}
  185. blob1.Size = int64(len(blob1.Data))
  186. blob2.Size = int64(len(blob2.Data))
  187. ra := fixtureRenameAnalysis()
  188. result, err := ra.blobsAreClose(blob1, blob2)
  189. assert.Nil(t, err)
  190. assert.True(t, result)
  191. blob1.Data = []byte("hello, mloncode")
  192. blob1.Size = int64(len(blob1.Data))
  193. result, err = ra.blobsAreClose(blob1, blob2)
  194. assert.Nil(t, err)
  195. assert.False(t, result)
  196. }
  197. func TestBlobsAreCloseBinary(t *testing.T) {
  198. blob1 := &CachedBlob{}
  199. blob2 := &CachedBlob{}
  200. ra := fixtureRenameAnalysis()
  201. result, err := ra.blobsAreClose(blob1, blob2)
  202. assert.Nil(t, err)
  203. assert.True(t, result)
  204. blob1.Data = make([]byte, 100)
  205. blob2.Data = blob1.Data
  206. blob1.Size = 100
  207. blob2.Size = 100
  208. result, err = ra.blobsAreClose(blob1, blob2)
  209. assert.Nil(t, err)
  210. assert.True(t, result)
  211. blob1.Data = bytes.Repeat([]byte{1}, 100)
  212. blob2.Data = blob1.Data
  213. result, err = ra.blobsAreClose(blob1, blob2)
  214. assert.Nil(t, err)
  215. assert.True(t, result)
  216. for i := 0; i < 100; i++ {
  217. blob1.Data[i] = byte(i)
  218. }
  219. result, err = ra.blobsAreClose(blob1, blob2)
  220. assert.Nil(t, err)
  221. assert.True(t, result)
  222. blob2.Data = make([]byte, 100)
  223. for i := 0; i < 80; i++ {
  224. blob2.Data[i] = byte(i)
  225. }
  226. result, err = ra.blobsAreClose(blob1, blob2)
  227. assert.Nil(t, err)
  228. assert.True(t, result)
  229. blob2.Data[79] = 0
  230. result, err = ra.blobsAreClose(blob1, blob2)
  231. assert.Nil(t, err)
  232. assert.False(t, result)
  233. blob1.Data = []byte("hello, world!")
  234. blob1.Size = int64(len(blob2.Data))
  235. result, err = ra.blobsAreClose(blob1, blob2)
  236. assert.Nil(t, err)
  237. assert.False(t, result)
  238. blob1.Data, blob2.Data = blob2.Data, blob1.Data
  239. blob1.Size, blob2.Size = blob2.Size, blob1.Size
  240. result, err = ra.blobsAreClose(blob1, blob2)
  241. assert.Nil(t, err)
  242. assert.False(t, result)
  243. }
  244. func loadData(t *testing.T, name string) []byte {
  245. gzsource, err := os.Open(path.Join("..", "test_data", name))
  246. defer gzsource.Close()
  247. if err != nil {
  248. t.Errorf("open ../test_data/%s: %v", name, err)
  249. }
  250. gzreader, err := gzip.NewReader(gzsource)
  251. if err != nil {
  252. t.Errorf("gzip ../test_data/%s: %v", name, err)
  253. }
  254. data, err := ioutil.ReadAll(gzreader)
  255. if err != nil {
  256. t.Errorf("gzip ../test_data/%s: %v", name, err)
  257. }
  258. return data
  259. }
  260. func TestBlobsAreCloseBug1(t *testing.T) {
  261. data := loadData(t, "rename_bug1.xml.gz")
  262. blob1 := &CachedBlob{Data: data}
  263. blob2 := &CachedBlob{Data: data}
  264. blob1.Size = int64(len(data))
  265. blob2.Size = int64(len(data))
  266. ra := fixtureRenameAnalysis()
  267. result, err := ra.blobsAreClose(blob1, blob2)
  268. assert.Nil(t, err)
  269. assert.True(t, result)
  270. }
  271. func TestBlobsAreCloseBug2(t *testing.T) {
  272. data1 := loadData(t, "rename_bug2.base.gz")
  273. data2 := loadData(t, "rename_bug2.head.gz")
  274. blob1 := &CachedBlob{Data: data1}
  275. blob2 := &CachedBlob{Data: data2}
  276. blob1.Size = int64(len(data1))
  277. blob2.Size = int64(len(data2))
  278. ra := fixtureRenameAnalysis()
  279. result, err := ra.blobsAreClose(blob1, blob2)
  280. assert.Nil(t, err)
  281. assert.False(t, result)
  282. }